Fine tropical cyclone rainstorm combination disaster risk assessment method

Abstract

The application discloses a fine tropical cyclone rainstorm combined disaster risk assessment method and belongs to the technical field of meteorological disaster assessment. The method comprises the following steps: obtaining multi-source disaster-causing data and performing pretreatment, extracting basic wind and rain indexes in historical cyclone events, calculating the cumulative occurrence probability of disaster-causing factors in different intensity intervals, and constructing a single-factor risk assessment index; determining the combination weight proportion of the precipitation factor and the gale factor; establishing multiple sets of risk grade division standards based on different spatial scales, generating historical risk zoning results; calculating the grid-based comprehensive risk value, and outputting the fine risk grade assessment map. The application solves the problems of the prior art, such as the lack of township scale risk identification, the deviation of the general weight from the local actual situation, and the inability to perform early pre-evaluation based on the forecast data, by fusing high-resolution grid data, calculating the localized wind and rain weight through regression, and substituting the grid forecast data into the evaluation model.

Description

Technical Field

[0001] This invention relates to the field of meteorological disaster assessment technology, and in particular to a refined method for assessing the risk of tropical cyclone wind and rain combination disasters. Background Technology

[0002] my country's eastern coastal areas are affected by multiple typhoons every year. Accurately assessing the disaster risk caused by typhoon wind and rain combinations is a key basis for emergency management departments to deploy disaster prevention forces in advance and organize the evacuation of people.

[0003] Currently, meteorological departments generally use spatial interpolation methods based on observational data from national meteorological stations for assessment. However, existing technologies have the following prominent problems: First, most counties and districts only have one national meteorological station, and townships and villages lack direct observational data. The interpolation results cannot reflect the differences in wind and rain across different underlying surfaces such as mountainous and coastal areas. Emergency departments can only obtain a rough conclusion of "high risk in a certain county" without being able to pinpoint specific townships or villages. Second, existing systems directly apply the nationally unified wind and rain weighting coefficients (wind 0.4, rain 0.6). However, in northern coastal provinces, typhoon disasters are mainly characterized by torrential rain and flooding. The general weights deviate significantly from local realities, leading to the incorrect downgrading of high-risk areas and misdirection of defense strategies. Third, existing technologies can only conduct assessments based on real-time data after a typhoon has passed. They cannot integrate numerical weather prediction data to generate gridded pre-assessment results before a typhoon arrives. Warnings can only be issued broadly at the county level, and can only be issued passively after the wind and rain have arrived, missing the golden time for proactive risk avoidance. Summary of the Invention

[0004] This application provides a refined method for assessing the risk of tropical cyclone wind and rain combinations, which solves the problems in the prior art such as the lack of risk identification at the township level, the deviation between general weights and local realities, and the inability to conduct advance assessments based on forecast data. It achieves the technical effects of precise risk positioning at the kilometer-level grid, localized weight adaptation, and proactive early warning before the arrival of typhoons.

[0005] This application provides a refined method for assessing the risk of tropical cyclone wind and rain combination disasters, including: acquiring multi-source disaster-causing data and preprocessing it, extracting basic wind and rain indicators from historical cyclone events, and normalizing disaster loss data to eliminate the impact of economic development background.

[0006] By using information diffusion processing logic, the limited observation samples are transformed into fuzzy sets, the cumulative occurrence probability of disaster-causing factors in different intensity ranges is calculated, and a single-factor hazard assessment index is constructed.

[0007] By analyzing the correlation between historical disaster loss data and various disaster-causing factors, the combined weight ratio of precipitation and strong wind factors is determined.

[0008] By combining real-time observation data with high-resolution simulated grid data, multiple sets of hazard level classification standards based on different spatial scales are established, and historical hazard zoning results are generated.

[0009] Substitute the real-time monitoring data or grid forecast data of the period to be evaluated into the preset hazard assessment model to calculate the gridded comprehensive hazard value;

[0010] Match the corresponding hazard level classification standards and output a refined risk level assessment map with spatial positioning information.

[0011] Furthermore, the steps for acquiring and preprocessing multi-source disaster data include:

[0012] Collect basic characteristic data including the location of the cyclone center, the maximum wind force at the center, and the movement path, and at the same time obtain daily precipitation, short-term heavy precipitation, and maximum wind speed data of all observation stations in the target area;

[0013] By retrieving historical disaster loss records and converting direct economic losses into the regional GDP of the target year, a relative loss coefficient that can characterize the severity of the disaster is obtained.

[0014] For each observation station, the average value and fluctuation range of each disaster-causing factor are calculated. The difference between the average value and one time the fluctuation range is set as the evaluation starting point, and invalid samples that do not reach this starting point are eliminated.

[0015] The pre-generated kilometer-level gridded wind field simulation data and gridded precipitation data are converted into different formats, and the gridded statistical values ​​corresponding to the observation station indicators are extracted to form a standardized input dataset covering the entire region.

[0016] Furthermore, the steps for calculating the cumulative occurrence probability using information diffusion processing logic include:

[0017] Based on the preset classification standards, the intensity of strong winds, the total rainfall during the process, and the maximum daily rainfall are divided into multiple increasing intensity ranges.

[0018] To address the problem of discontinuous probability distribution caused by insufficient historical sample size, a diffusion function is introduced to diffuse energy from each discrete observation to the surrounding intensity range.

[0019] Based on the total diffusion energy received in each intensity interval, the fuzzy membership degree corresponding to that interval is calculated, thereby constructing the probability distribution function of the disaster-causing factor;

[0020] The probability values ​​of each intensity range are summed to obtain the cumulative probability of different disaster-causing factors in each level range, and this is used as the basis for measuring the degree of danger of a single disaster-causing factor.

[0021] Further steps to determine the combined weighting of precipitation and wind factors include:

[0022] A mathematical correlation model was established with the normalized relative loss coefficient as the target variable and the maximum wind speed, process rainfall, and daily maximum rainfall as independent variables.

[0023] Identify and remove anomalous samples from historical data whose loss values ​​deviate significantly from the normal range to ensure that the weight calculation results can objectively reflect the characteristics of conventional disasters;

[0024] Regression analysis was used to calculate the contribution of the precipitation index set and the wind index set to the final loss, and preliminary proportional values ​​were obtained.

[0025] By comparing two typical cases of wind-induced disasters and rain-induced disasters, the average loss intensity corresponding to each case was calculated and weighted by the values ​​obtained from regression analysis to finally determine the weights of precipitation and wind factors.

[0026] Furthermore, the steps to establish multiple sets of hazard level classification standards based on different spatial scales include:

[0027] The comprehensive hazard values ​​of each station in the historical cyclone events were calculated and summarized to form a sample pool reflecting the regional hazard level.

[0028] Calculate the average level and dispersion index of the data in the sample pool, and classify the risk into four levels: high risk, relatively high risk, relatively low risk, and low risk according to the preset multiple relationship;

[0029] To address the characteristics of different data sources, three differentiated grading standards were established: one based on representative observation stations, one based on observation stations across the entire region, and one based on high-resolution simulated grid points.

[0030] Each standard includes specific numerical range limits, which are used to accurately map the evaluation object to a grade under different data precision conditions.

[0031] Further steps in calculating the gridded composite hazard value include:

[0032] When assessing a single cyclone event, multiple precipitation indicators are extracted, including the maximum precipitation over three consecutive hours, the maximum daily precipitation, and the cumulative precipitation over the event, as well as the maximum wind speed indicator.

[0033] Set a trigger threshold for each indicator. For measured values ​​that exceed the threshold, use a normalization algorithm to map them to a dimensionless space between zero and one.

[0034] Calculate the average of the normalized values ​​of multiple precipitation indices and use it as the comprehensive precipitation hazard component for that grid point;

[0035] The comprehensive precipitation hazard component and the normalized wind force value are multiplied by their respective combined weight ratios and summed to obtain the final comprehensive hazard value of the grid point under the influence of this cyclone.

[0036] Furthermore, the steps for outputting a refined risk level assessment map with spatial positioning information include:

[0037] Based on the accuracy of the data used in the current assessment, a matching target threshold system is selected from three preset grading standards;

[0038] Traverse all geographical units within the target area, compare the comprehensive hazard value of each kilometer-level grid point with the target threshold system, and determine its hazard level.

[0039] By associating the hazard level information with coordinate data in the geographic information system, using colors to distinguish different hazard levels, a detailed assessment distribution map of the entire region is drawn.

[0040] By overlaying charts, specific townships or villages covered by high-risk levels are identified, generating decision-making reference data to guide resource allocation and personnel relocation.

[0041] Furthermore, the steps of substituting grid forecast data into a pre-defined hazard assessment model include:

[0042] Acquire gridded weather forecast products for future time periods generated by numerical weather prediction systems, which have a kilometer-level spatial resolution consistent with actual assessments;

[0043] The hourly wind speed variation sequence and precipitation accumulation sequence are extracted from the forecast products, and the maximum precipitation of the forecast day, the total rainfall during the forecast process, and the maximum wind speed are further calculated.

[0044] The forecast indicators are filtered at the starting point and normalized using a method consistent with the logic of real-time data processing.

[0045] Before the cyclone actually arrives, the comprehensive risk trends that may occur in the future are calculated based on forecast data, so as to identify and warn of potential disaster areas in advance.

[0046] Further steps in generating historical hazard zoning results include:

[0047] The assessment data of all historical cyclone cases in the target area are summarized to calculate the frequency and average hazard intensity of wind and rain disasters suffered by each geographical unit over a long period.

[0048] Historical distribution maps of single factors, primarily influenced by wind and primarily influenced by precipitation, were drawn separately to identify differences in regional disaster-causing characteristics.

[0049] A background risk zoning base map covering the entire province is generated by combining comprehensive hazard values. This base map serves as a calibration reference for the results of a single cyclone assessment.

[0050] The assessment results are integrated into the disaster census calculation module. By comparing and verifying the results with the actual distribution of disaster sites, the weight parameters in the assessment model are dynamically optimized.

[0051] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0052] By introducing high-resolution gridded meteorological data and standardized preprocessing of multi-source data, the traditional extensive assessment relying on sparse station interpolation is upgraded to a fine-grained scale covering every township and village. Emergency management departments can no longer only obtain vague conclusions at the district and county level, but can intuitively grasp accurate information such as "which township will experience the most severe wind and rain, and which village needs to be evacuated in advance," thus solving the problem of lack of risk identification at the grassroots level.

[0053] Furthermore, based on the detailed assessment results, by constructing multiple sets of hazard level classification standards applicable to different data accuracies, and combining them with gridded risk level assessment maps output by a geographic information system, it is possible to pinpoint specific villages and settlements covered by high-risk levels. Emergency material reserves and rescue forces can then be deployed proactively, like "following the map," ensuring sufficient resources in high-risk areas and preventing idleness and waste in low-risk areas, thus reducing the misallocation of resources in the traditional approach of spreading resources evenly.

[0054] Furthermore, in the process of achieving precise resource deployment, grid forecast data for future periods can be substituted into the evaluation model to calculate the possible comprehensive risk trends two days before the typhoon actually makes landfall, generating an early warning risk distribution map, transforming passive disaster relief into proactive risk avoidance, and improving the overall effectiveness of typhoon disaster prevention. Attached Figure Description

[0055] Figure 1 A flowchart illustrating the refined tropical cyclone storm combination hazard risk assessment method provided in this application embodiment. Detailed Implementation

[0056] This application provides a refined method for assessing the risk of tropical cyclone wind and rain combinations, which solves the problems in the prior art such as the lack of risk identification at the township level, the deviation of general weights from local realities, and the inability to conduct advance assessments based on forecast data. By integrating high-resolution grid data, regressing to calculate localized wind and rain weights, and substituting grid forecast data into the assessment model, it achieves the technical effects of accurate risk positioning at the kilometer-level grid level, localized weight adaptation, and proactive early warning before typhoon landfall.

[0057] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0058] like Figure 1 The diagram shows a flowchart of a refined tropical cyclone wind and rain combination disaster risk assessment method provided in this application embodiment. The method includes the following steps: acquiring multi-source disaster-causing data and preprocessing it, extracting basic wind and rain indicators from historical cyclone events, and normalizing disaster loss data to eliminate the impact of economic development background.

[0059] By using information diffusion processing logic, the limited observation samples are transformed into fuzzy sets, the cumulative occurrence probability of disaster-causing factors in different intensity ranges is calculated, and a single-factor hazard assessment index is constructed.

[0060] By analyzing the correlation between historical disaster loss data and various disaster-causing factors, the combined weight ratio of precipitation and strong wind factors is determined.

[0061] By combining real-time observation data with high-resolution simulated grid data, multiple sets of hazard level classification standards based on different spatial scales are established, and historical hazard zoning results are generated.

[0062] Substitute the real-time monitoring data or grid forecast data of the period to be evaluated into the preset hazard assessment model to calculate the gridded comprehensive hazard value;

[0063] Match the corresponding hazard level classification standards and output a refined risk level assessment map with spatial positioning information.

[0064] In this embodiment, raw information on the impact of typhoons is collected through multiple channels. This information includes not only monitoring values ​​of the weather itself, but also records of socio-economic losses caused by the weather. During the data transfer process, this messy data is first cleaned and standardized to ensure the comparability of data from different years and of different natures.

[0065] Next, by using specific fuzzy processing techniques, the problem of inaccurate data statistics caused by the limited number of historical typhoon cases is solved, thereby scientifically measuring the disaster potential of each individual meteorological factor (such as wind speed and rainfall).

[0066] Subsequently, by establishing a mathematical correlation between losses and meteorological indicators, the method automatically allocates the "voice" ratio of wind and rain in the total risk. Finally, this method applies these logics to fine-grained grid cells, transforming abstract hazard values ​​into intuitive hazard level maps by comparing them with historical risk scales, thereby capturing disaster risks at the village and town level.

[0067] Furthermore, the steps for acquiring and preprocessing multi-source disaster data include:

[0068] Collect basic characteristic data including the location of the cyclone center, the maximum wind force at the center, and the movement path, and at the same time obtain daily precipitation, short-term heavy precipitation, and maximum wind speed data of all observation stations in the target area;

[0069] By retrieving historical disaster loss records and converting direct economic losses into the regional GDP of the target year, a relative loss coefficient that can characterize the severity of the disaster is obtained.

[0070] For each observation station, the average value and fluctuation range of each disaster-causing factor are calculated. The difference between the average value and one time the fluctuation range is set as the evaluation starting point, and invalid samples that do not reach this starting point are eliminated.

[0071] The pre-generated kilometer-level gridded wind field simulation data and gridded precipitation data are converted into different formats, and the gridded statistical values ​​corresponding to the observation station indicators are extracted to form a standardized input dataset covering the entire region.

[0072] In this embodiment, during the step of acquiring and preprocessing multi-source data, the optimal path dataset of tropical cyclones is automatically retrieved through a computer interface, and key information such as cyclone number, central pressure, maximum wind speed, and landfall location is extracted.

[0073] The meteorological monitoring data is further divided into two parts: one is daily and hourly data provided by fixed observation stations distributed in various counties and cities; the other is kilometer-level simulated grid data covering the entire province, generated based on satellite and radar signals.

[0074] To mitigate the disruptions caused by rising prices and economic growth, this embodiment uses a relative loss rate for calculation. The calculation formula is as follows:

[0075] ;

[0076] in, Represents the relative loss coefficient. This represents the absolute value of the direct economic losses caused by a particular typhoon. This indicates the region's GDP for the year the typhoon occurred.

[0077] When setting the assessment starting point, the historical wind speed and rainfall data of each observation station are statistically analyzed, and the average value minus one standard deviation is taken as the "threshold". If the wind and rain values ​​brought by a typhoon are lower than this threshold, it is judged as "not hazardous" and its weight is reset to zero in subsequent calculations, thereby avoiding interference from minor rainfall or ordinary gusts on the disaster assessment results.

[0078] Furthermore, the steps for calculating the cumulative occurrence probability using information diffusion processing logic include:

[0079] Based on the preset classification standards, the intensity of strong winds, the total rainfall during the process, and the maximum daily rainfall are divided into multiple increasing intensity ranges.

[0080] To address the problem of discontinuous probability distribution caused by insufficient historical sample size, a diffusion function is introduced to diffuse energy from each discrete observation to the surrounding intensity range.

[0081] Based on the total diffusion energy received in each intensity interval, the fuzzy membership degree corresponding to that interval is calculated, thereby constructing the probability distribution function of the disaster-causing factor;

[0082] The probability values ​​of each intensity range are summed to obtain the cumulative probability of different disaster-causing factors in each level range, and this is used as the basis for measuring the degree of danger of a single disaster-causing factor.

[0083] In this embodiment, when constructing the index using information diffusion processing logic, considering that typhoons are extreme weather events with limited historical samples, directly calculating the probability can easily lead to jumps and biases. The following smoothing logic is used:

[0084] First, a set of discrete intensity level points (e.g., wind speed levels) is defined. For each actually observed sample value... Instead of simply assigning it to a single point, it uses a diffusion function to distribute its energy to surrounding level points. :

[0085] ;

[0086] in, For the first The actual observed values ​​of each sample For the preset intensity level points, The diffusion coefficient is determined based on the maximum and minimum values ​​of the sample and the total sample size.

[0087] In this way, the originally isolated data points are transformed into an "energy cloud" covering a certain area, with all samples at a certain level. After the energy is superimposed, dividing by the total energy yields the probability of occurrence for that level. Finally, these probabilities are summed level by level to form a cumulative probability distribution. This process ensures that even with a very small sample size, the assessment model can still produce stable and continuous hazard indicators.

[0088] Further steps to determine the combined weighting of precipitation and wind factors include:

[0089] A mathematical correlation model was established with the normalized relative loss coefficient as the target variable and the maximum wind speed, process rainfall, and daily maximum rainfall as independent variables.

[0090] Identify and remove anomalous samples from historical data whose loss values ​​deviate significantly from the normal range to ensure that the weight calculation results can objectively reflect the characteristics of conventional disasters;

[0091] Regression analysis was used to calculate the contribution of the precipitation index set and the wind index set to the final loss, and preliminary proportional values ​​were obtained.

[0092] By comparing two typical cases of wind-induced disasters and rain-induced disasters, the average loss intensity corresponding to each case was calculated and weighted by the values ​​obtained from regression analysis to finally determine the weights of precipitation and wind factors.

[0093] In this embodiment, the process of determining the weights of the wind and rain combination aims to identify which component, wind or rain, contributes more to the disaster. Logical modeling is performed using multiple linear regression.

[0094] ;

[0095] in, The relative loss coefficient calculated above, The total precipitation during the process, This is the highest daily rainfall. Let be the maximum wind speed, and a, b, and c be the regression coefficients to be determined. This is the error term.

[0096] In practice, by automatically scanning the historical database, extreme outlier cases such as "abnormally high proportion of disaster damage" are identified and removed to prevent the model from being biased by a few extreme cases. After calculating the coefficients, the precipitation-related coefficients ( The coefficient related to wind force ( The weights are normalized to obtain the final weights. For example, in the practical application of this embodiment, after simulating dozens of historical typhoons, the weight allocation of the rain factor is finally determined to be 0.56, and the weight allocation of the wind factor is 0.44.

[0097] Furthermore, the steps to establish multiple sets of hazard level classification standards based on different spatial scales include:

[0098] The comprehensive hazard values ​​of each station in the historical cyclone events were calculated and summarized to form a sample pool reflecting the regional hazard level.

[0099] Calculate the average level and dispersion index of the data in the sample pool, and classify the risk into four levels: high risk, relatively high risk, relatively low risk, and low risk according to the preset multiple relationship;

[0100] To address the characteristics of different data sources, three differentiated grading standards were established: one based on representative observation stations, one based on observation stations across the entire region, and one based on high-resolution simulated grid points.

[0101] Each standard includes specific numerical range limits, which are used to accurately map the evaluation object to a grade under different data precision conditions.

[0102] In this embodiment, when establishing the hazard level classification standard, three interrelated scales are preset to adapt to different application scenarios (such as preliminary estimation and precise review):

[0103] The first set is the "Local Historical Scale," which is mainly used to assess the extreme levels of a single observation station; the second set is the "Regional Historical Scale," which mixes the historical extreme values ​​of all stations to measure the level of the area within the province; the third set is the "Grid Simulation Scale," which is specifically adapted for simulation data with a 1-kilometer resolution.

[0104] The specific logic for classifying the levels is: calculate the average value of the historical sample set. and standard deviation .

[0105] High risk level: numerical value ;

[0106] Higher risk level: numerical values ;

[0107] Lower risk level: numerical values ;

[0108] Low risk level: numerical value .

[0109] This classification method does not rely on subjective experience, but is automatically generated based entirely on the distribution characteristics of the data, ensuring the objectivity of the evaluation results, and can automatically evolve as the historical database is updated.

[0110] Further steps in calculating the gridded composite hazard value include:

[0111] When assessing a single cyclone event, multiple precipitation indicators are extracted, including the maximum precipitation over three consecutive hours, the maximum daily precipitation, and the cumulative precipitation over the event, as well as the maximum wind speed indicator.

[0112] Set a trigger threshold for each indicator. For measured values ​​that exceed the threshold, use a normalization algorithm to map them to a dimensionless space between zero and one.

[0113] Calculate the average of the normalized values ​​of multiple precipitation indices and use it as the comprehensive precipitation hazard component for that grid point;

[0114] The comprehensive precipitation hazard component and the normalized wind force value are multiplied by their respective combined weight ratios and summed to obtain the final comprehensive hazard value of the grid point under the influence of this cyclone.

[0115] In this embodiment, a refined time-dimensional index is introduced for the comprehensive hazard calculation of a single typhoon event. The calculation steps are as follows:

[0116] First, extract four key features for each grid point in this typhoon: total rainfall, maximum daily rainfall, maximum rainfall over three consecutive hours, and maximum daily wind speed.

[0117] To enable data from different units to be calculated together, a normalization operation is performed:

[0118] ;

[0119] in, The value is the normalized value. These are actual observed or predicted values. As the aforementioned starting point for evaluation, This is the historical maximum value of the indicator. If the actual value is lower than the starting point, then... Take 0 directly.

[0120] Finally, the normalized values ​​of the three rainfall indicators are averaged to obtain the "comprehensive rainfall index," which is then weighted and summed with the "wind index" according to weights of 0.56 and 0.44. This design fully considers the sudden contribution of short-duration heavy rainfall (three-hour rainfall) to the occurrence of flash floods and urban flooding, and is more scientific than the traditional method that only considers total rainfall.

[0121] Furthermore, the steps for outputting a refined risk level assessment map with spatial positioning information include:

[0122] Based on the accuracy of the data used in the current assessment, a matching target threshold system is selected from three preset grading standards;

[0123] Traverse all geographical units within the target area, compare the comprehensive hazard value of each kilometer-level grid point with the target threshold system, and determine its hazard level.

[0124] By associating the hazard level information with coordinate data in the geographic information system, using colors to distinguish different hazard levels, a detailed assessment distribution map of the entire region is drawn.

[0125] By overlaying charts, specific townships or villages covered by high-risk levels are identified, generating decision-making reference data to guide resource allocation and personnel relocation.

[0126] In this embodiment, the process of outputting a refined evaluation map is key to making abstract numerical values ​​concrete. Based on the calculation results of each 1km×1km grid point, the corresponding color code is automatically matched (e.g., red represents high risk, and blue represents no risk).

[0127] During execution, spatial geographic coordinate information is used to vector-overlay the calculated grid data with the administrative division base map. Through this logical association, the geofences covered by high-risk grid points can be automatically identified, and the names of towns and villages within those geofences can be retrieved in reverse.

[0128] Furthermore, this embodiment supports dynamic adjustment of the display threshold. For example, when the emergency response level is upgraded, managers can lower the trigger threshold for "high risk," thereby expanding the coverage of the early warning. This configuration method makes the evaluation map produced by this invention not only a static display, but also a map that can directly guide the "targeted deployment" of rescue forces.

[0129] Furthermore, the steps of substituting grid forecast data into a pre-defined hazard assessment model include:

[0130] Acquire gridded weather forecast products for future time periods generated by numerical weather prediction systems, which have a kilometer-level spatial resolution consistent with actual assessments;

[0131] The hourly wind speed variation sequence and precipitation accumulation sequence are extracted from the forecast products, and the maximum precipitation of the forecast day, the total rainfall during the forecast process, and the maximum wind speed are further calculated.

[0132] The forecast indicators are filtered at the starting point and normalized using a method consistent with the logic of real-time data processing.

[0133] Before the cyclone actually arrives, the comprehensive risk trends that may occur in the future are calculated based on forecast data, so as to identify and warn of potential disaster areas in advance.

[0134] Further steps in generating historical hazard zoning results include:

[0135] The assessment data of all historical cyclone cases in the target area are summarized to calculate the frequency and average hazard intensity of wind and rain disasters suffered by each geographical unit over a long period.

[0136] Historical distribution maps of single factors, primarily influenced by wind and primarily influenced by precipitation, were drawn separately to identify differences in regional disaster-causing characteristics.

[0137] A background risk zoning base map covering the entire province is generated by combining comprehensive hazard values. This base map serves as a calibration reference for the results of a single cyclone assessment.

[0138] The assessment results are integrated into the disaster census calculation module. By comparing and verifying the results with the actual distribution of disaster sites, the weight parameters in the assessment model are dynamically optimized.

[0139] In this embodiment, generating historical hazard zoning results and operational application steps serves as a closed-loop verification of the effectiveness of the assessment model. By accumulating the assessment results of all typhoons over the past few decades along a timeline, it identifies which areas are "disaster hotspots" that are constantly plagued by wind and rain.

[0140] In practical applications, the overlap between historical assessment maps and actual post-disaster loss distribution maps is automatically calculated. If a region is found to have repeatedly shown "low assessed risk but large actual losses," a weight adjustment mechanism is triggered, automatically increasing the weight coefficient of specific disaster-causing factors in that region.

[0141] Furthermore, this achievement, through a standardized data interface, is directly embedded into the meteorological disaster risk survey module, allowing county-level users to directly query the hazard background of each village within their jurisdiction via a browser. This "cloud computing, local application" model lowers the technical threshold for small and medium-sized meteorological service units, ensuring the ease of implementation and wide applicability of the solution.

[0142] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0143] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0144] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0145] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0146] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0147] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A refined method for assessing the hazard risk of tropical cyclone wind and rain combinations, characterized in that, Includes the following steps: Acquire and preprocess multi-source disaster data, extract basic wind and rain indicators from historical cyclone events, and normalize disaster loss data to eliminate the impact of economic development background. By using information diffusion processing logic, the limited observation samples are transformed into fuzzy sets, the cumulative occurrence probability of disaster-causing factors in different intensity ranges is calculated, and a single-factor hazard assessment index is constructed. By analyzing the correlation between historical disaster loss data and various disaster-causing factors, the combined weight ratio of precipitation and strong wind factors is determined. By combining real-time observation data with high-resolution simulated grid data, multiple sets of hazard level classification standards based on different spatial scales are established, and historical hazard zoning results are generated. Substitute the real-time monitoring data or grid forecast data of the period to be evaluated into the preset hazard assessment model to calculate the gridded comprehensive hazard value; Match the corresponding hazard level classification standards and output a refined risk level assessment map with spatial positioning information.

2. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps for acquiring and preprocessing multi-source disaster data include: Collect basic characteristic data including the location of the cyclone center, the maximum wind force at the center, and the movement path, and at the same time obtain daily precipitation, short-term heavy precipitation, and maximum wind speed data of all observation stations in the target area; By retrieving historical disaster loss records and converting direct economic losses into the regional GDP of the target year, a relative loss coefficient that can characterize the severity of the disaster is obtained. For each observation station, the average value and fluctuation range of each disaster-causing factor are calculated. The difference between the average value and one time the fluctuation range is set as the evaluation starting point, and invalid samples that do not reach this starting point are eliminated. The pre-generated kilometer-level gridded wind field simulation data and gridded precipitation data are converted into different formats, and the gridded statistical values ​​corresponding to the observation station indicators are extracted to form a standardized input dataset covering the entire region.

3. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps for calculating the cumulative probability of occurrence using information diffusion processing logic include: Based on the preset classification standards, the intensity of strong winds, the total rainfall during the process, and the maximum daily rainfall are divided into multiple increasing intensity ranges. To address the problem of discontinuous probability distribution caused by insufficient historical sample size, a diffusion function is introduced to diffuse energy from each discrete observation to the surrounding intensity range. Based on the total diffusion energy received in each intensity interval, the fuzzy membership degree corresponding to that interval is calculated, thereby constructing the probability distribution function of the disaster-causing factor; The probability values ​​of each intensity range are summed to obtain the cumulative probability of different disaster-causing factors in each level range, and this is used as the basis for measuring the degree of danger of a single disaster-causing factor.

4. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps to determine the combined weighting of precipitation and wind factors include: A mathematical correlation model was established with the normalized relative loss coefficient as the target variable and the maximum wind speed, process rainfall, and daily maximum rainfall as independent variables. Identify and remove anomalous samples from historical data whose loss values ​​deviate significantly from the normal range to ensure that the weight calculation results can objectively reflect the characteristics of conventional disasters; Regression analysis was used to calculate the contribution of the precipitation index set and the wind index set to the final loss, and preliminary proportional values ​​were obtained. By comparing two typical cases, one dominated by wind disasters and the other by rain disasters, the average loss intensity was calculated for each case, and then weighted and averaged with the values ​​obtained from regression analysis to finally determine the weights of precipitation and wind factors.

5. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps to establish multiple sets of hazard level classification standards based on different spatial scales include: The comprehensive hazard values ​​of each station in the historical cyclone events were calculated and summarized to form a sample pool reflecting the regional hazard level. Calculate the average level and dispersion index of the data in the sample pool, and classify the risk into four levels: high risk, relatively high risk, relatively low risk, and low risk according to the preset multiple relationship; To address the characteristics of different data sources, three differentiated grading standards were established: one based on representative observation stations, one based on observation stations across the entire region, and one based on high-resolution simulated grid points. Each standard includes specific numerical range limits, which are used to accurately map the evaluation object to a grade under different data precision conditions.

6. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps for calculating the gridded composite hazard value include: When assessing a single cyclone event, multiple precipitation indicators are extracted, including the maximum precipitation over three consecutive hours, the maximum daily precipitation, and the cumulative precipitation over the event, as well as the maximum wind speed indicator. Set a trigger threshold for each indicator. For measured values ​​that exceed the threshold, use a normalization algorithm to map them to a dimensionless space between zero and one. Calculate the average of the normalized values ​​of multiple precipitation indices and use it as the comprehensive precipitation hazard component for that grid point; The comprehensive precipitation hazard component and the normalized wind force value are multiplied by their respective combined weight ratios and summed to obtain the final comprehensive hazard value of the grid point under the influence of this cyclone.

7. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps for generating a refined risk level assessment map with spatial positioning information include: Based on the accuracy of the data used in the current assessment, a matching target threshold system is selected from three preset grading standards; Traverse all geographical units within the target area, compare the comprehensive hazard value of each kilometer-level grid point with the target threshold system, and determine its hazard level. By associating the hazard level information with coordinate data in the geographic information system, using colors to distinguish different hazard levels, a detailed assessment distribution map of the entire region is drawn. By overlaying charts, specific townships or villages covered by high-risk levels are identified, generating decision-making reference data to guide resource allocation and personnel relocation.

8. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps for substituting grid forecast data into a pre-defined hazard assessment model include: Acquire gridded weather forecast products for future time periods generated by numerical weather prediction systems, which have a kilometer-level spatial resolution consistent with actual assessments; The hourly wind speed variation sequence and precipitation accumulation sequence are extracted from the forecast products, and the maximum precipitation of the forecast day, the total rainfall during the forecast process, and the maximum wind speed are further calculated. The forecast indicators are filtered at the starting point and normalized using a method consistent with the logic of real-time data processing. Before the cyclone actually arrives, the comprehensive risk trends that may occur in the future are calculated based on forecast data, so as to identify and warn of potential disaster areas in advance.

9. The refined tropical cyclone wind and rain combination hazard risk assessment method as described in claim 1, characterized in that, The steps to generate historical hazard zoning results include: The assessment data of all historical cyclone cases in the target area are summarized to calculate the frequency and average hazard intensity of wind and rain disasters suffered by each geographical unit over a long period. Historical distribution maps of single factors, primarily influenced by wind and primarily influenced by precipitation, were drawn separately to identify differences in regional disaster-causing characteristics. A background risk zoning base map covering the entire province is generated by combining comprehensive hazard values. This base map serves as a calibration reference for the results of a single cyclone assessment. The assessment results are integrated into the disaster census calculation module. By comparing and verifying the results with the actual distribution of disaster sites, the weight parameters in the assessment model are dynamically optimized.

Images